simulation experiments under various scenarios. For space reasons, in
what follows we will present the results of two typical cases in the
priority queuing system with Hurst parameter Hf¼0.85 and service
capacity C¼120. The mean arrival rates lf and lp are set as follows:
Case I: lf¼90 and lp¼20; Case II: lf¼20 and lp¼90. Fig. 1 shows
that the simulation results of queue length distributions and loss
probabilities of FBM and Poisson traffic closely match their corresponding
analytical results for both cases. The results demonstrate
that the developed analytical model has a good degree of accuracy in
predicting the queue length distributions and loss probabilities of both
traffic flows. It is worth noting that the analytical and simulation
results of both FBM and Poisson traffic appear in Figs. 1a and b,
corresponding to Case I. However, in Figs. 1c and d, which correspond
to Case II, only the results of Poisson traffic are plotted. This is
because the arrival rate of the high priority FBM traffic in Case II is
much smaller than the system service capacity. As a result, FBM
traffic can be served in time and thus no packets wait for service in the
buffer and no packets are lost due to buffer overflow.